Why drift costs more than it looks
A dispenser that delivers slightly more than it charges for is giving fuel away continuously, and nothing about the transaction looks wrong. The customer gets what they paid for and a little more. The totaliser records what was actually dispensed. The cash matches the litres the meter counted. Every record is internally consistent.
The only place it shows is in stock: the tank empties faster than the sales say it should. And because it is proportional to volume, a busy nozzle bleeds far more than a quiet one — which is also what makes it identifiable.
It is worth being clear that drift can run the other way too. A dispenser that under-delivers is a legal problem rather than a commercial one, and a much more serious kind of trouble.
The signature that identifies it
Calibration drift is proportional to litres dispensed on one specific nozzle and absent on its neighbours drawing from the same tank. Nothing else on the list of causes has that shape — evaporation tracks temperature and total volume, a leak tracks time regardless of throughput. That is why per-nozzle records matter: a pump that only records tank-level movement can see the loss and never locate it.
Verification and stamping
Dispensing units at a fuel outlet are measuring instruments in the legal sense, and their verification is a statutory matter under India's weights and measures regime rather than a maintenance preference. Verification is carried out by the Legal Metrology department and the unit is stamped when it passes.
The specifics — who inspects, on what cycle, what the permitted tolerance is, and what the process costs — are set by rules that vary and change. Take those from your state's Legal Metrology department and from the terms of your dealership rather than from a web page.
What does not vary is the operational point: the stamp certifies the dispenser on the day it was tested. It says nothing about what the unit is doing eight months later, which is what your own daily records are for.
Your own five-litre check
Between verifications, the check most outlets run is a five-litre measure. It takes a couple of minutes per nozzle and it is the only routine that catches drift early.
- Use the measure supplied and verified for the purpose, not any five-litre container.
- Let the measure drain properly between tests — residue is a large fraction of a tolerance.
- Dispense at the normal flow rate. A slow trickle and a full-flow delivery are not the same test.
- Do it at the same time of day where you can; temperature affects the volume you are measuring.
- Record every test: nozzle, date, and the litres the measure actually held.
What testing litres do to your figures
This is the part that catches people out, and it matters more than the test itself.
Fuel pushed into a test measure leaves the nozzle, so the totaliser counts it — and then goes straight back into the tank. It was dispensed and it was not sold.
Counted as a sale, those litres inflate the revenue you expect and manufacture a cash shortfall of exactly their value. Ignored entirely, they reappear later as a stock gain nobody ordered. Recorded properly — deducted from litres sold, kept in litres dispensed — both the cash and the stock reconcile.
The meter reading calculator handles the deduction, and the daily sales report format gives testing litres their own column for exactly this reason.
Finding drift in records you already keep
If you keep meter readings per nozzle and a dip per tank, the analysis is a subtraction you can do on paper.
- Take a month of daily stock variance for one tank.
- Take the litres each nozzle on that tank dispensed over the same month.
- A tank whose variance scales with one nozzle's throughput, and does not move when that nozzle is quiet, has a drifting dispenser rather than a stock problem.
- A variance that appears on a low-volume day is much more interesting than one on a busy day — drift cannot produce a large loss on a day that moved almost nothing. That day points at something else.
When a dispenser is replaced
A replacement unit starts its totaliser from a new number, and this is the one situation where a reading can legitimately appear to run backwards.
Record it as two reading pairs, not one: close the old nozzle at its final figure and open the new one at whatever it starts from. Treating the changeover as a single pair produces either a phantom sale or a phantom loss the size of the difference between the two meters — a number large enough to send someone looking for a leak that does not exist.
Frequently asked questions
How often should I test my nozzles myself?
Often enough that drift shows up as a small number rather than a month of losses — many outlets do it weekly per nozzle, and the marginal cost is a few minutes. Statutory verification is separate and runs on the cycle your state's Legal Metrology rules set.
What tolerance is acceptable?
The permitted tolerance is set by law rather than by preference, and you should take it from your state's Legal Metrology department. Operationally, what you are watching for is a *change* — a nozzle that has moved relative to its own history is telling you something whether or not it is still inside tolerance.
Should testing litres be counted as a sale?
No. They left the nozzle so the totaliser counted them, and they went straight back into the tank. Deduct them from litres sold and keep them in litres dispensed, or you will manufacture a cash shortfall on the day and a stock gain later.
Can software detect a drifting nozzle?
Not on its own — a variance is a number, and telling drift from evaporation or a leak takes the shape of that number over weeks against per-nozzle throughput. What records per nozzle per shift do is make the analysis possible at all. A pump that only records tank-level movement can see the loss and never locate it.